U.S. patent number 10,743,789 [Application Number 15/739,719] was granted by the patent office on 2020-08-18 for ecg signal parallel analysis apparatus, method and mobile terminal.
This patent grant is currently assigned to SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY, CHINESE ACADEMY OF SCIENCES. The grantee listed for this patent is SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY, CHINESE ACADEMY OF SCIENCES. Invention is credited to Yunpeng Cai, Xiaomao Fan, Ye Li, Yujie Yang, Qihang Yao.
![](/patent/grant/10743789/US10743789-20200818-D00000.png)
![](/patent/grant/10743789/US10743789-20200818-D00001.png)
![](/patent/grant/10743789/US10743789-20200818-D00002.png)
![](/patent/grant/10743789/US10743789-20200818-D00003.png)
United States Patent |
10,743,789 |
Li , et al. |
August 18, 2020 |
ECG signal parallel analysis apparatus, method and mobile
terminal
Abstract
Provided are an electrocardiogram signal parallel analysis
apparatus, a mobile terminal incorporating the apparatus, and
related methods. The apparatus includes an integrated memory, a
central processing unit and a graphic processing unit. The
integrated memory includes a first memory and a second memory for
being used by the central processing unit and the graphic
processing unit respectively, and the central processing unit may
access the second memory. The central processing unit performs
primary noise reduction on a received electrocardiogram original
signal to obtain a primary electrocardiogram signal, and performs
abnormal heartbeat classification preliminary screening on
characteristic data extracted from the graphic processing unit to
obtain suspected abnormal heartbeat data. The graphic processing
unit performs characteristic extraction on the primary
electrocardiogram signal to obtain characteristic data, performs
secondary noise reduction on the primary electrocardiogram signal
to obtain a secondary electrocardiogram signal, and processes the
suspected abnormal heartbeat data and the secondary
electrocardiogram signal by applying a template matching
classification mode to obtain final abnormal heartbeat data.
Inventors: |
Li; Ye (Guangdong,
CN), Fan; Xiaomao (Guangdong, CN), Cai;
Yunpeng (Guangdong, CN), Yao; Qihang (Guangdong,
CN), Yang; Yujie (Guangdong, CN) |
Applicant: |
Name |
City |
State |
Country |
Type |
SHENZHEN INSTITUTES OF ADVANCED TECHNOLOGY, CHINESE ACADEMY OF
SCIENCES |
Shenzhen, Guangdong |
N/A |
CN |
|
|
Assignee: |
SHENZHEN INSTITUTES OF ADVANCED
TECHNOLOGY, CHINESE ACADEMY OF SCIENCES (Shenzhen,
CN)
|
Family
ID: |
66534679 |
Appl.
No.: |
15/739,719 |
Filed: |
November 28, 2017 |
PCT
Filed: |
November 28, 2017 |
PCT No.: |
PCT/CN2017/113408 |
371(c)(1),(2),(4) Date: |
December 23, 2017 |
PCT
Pub. No.: |
WO2019/100417 |
PCT
Pub. Date: |
May 31, 2019 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20190150778 A1 |
May 23, 2019 |
|
Foreign Application Priority Data
|
|
|
|
|
Nov 21, 2017 [CN] |
|
|
2017 1 1169664 |
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
A61B
5/35 (20210101); A61B 5/316 (20210101); A61B
5/332 (20210101); A61B 5/0022 (20130101); A61B
5/366 (20210101); G16H 10/60 (20180101); G16H
50/20 (20180101); A61B 5/352 (20210101); A61B
5/7203 (20130101); A61B 5/0006 (20130101); A61B
5/335 (20210101); A61B 5/0245 (20130101) |
Current International
Class: |
A61B
5/0452 (20060101); A61B 5/0404 (20060101); G16H
50/20 (20180101); A61B 5/0472 (20060101); A61B
5/00 (20060101); A61B 5/04 (20060101); A61B
5/0245 (20060101); A61B 5/0456 (20060101); A61B
5/0432 (20060101) |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Stice; Paula J
Claims
The invention claimed is:
1. An electrocardiogram signal parallel analysis apparatus, wherein
the apparatus comprises an integrated memory, a central processor
and a graphic processor, wherein the integrated memory is coupled
to the central processor and to the graphic processor and comprises
a first memory configured for being used by the central processor
and a second memory configured for being used by the graphic
processor, the central processor is configured to be able to access
the second memory, and the central processor and the graphic
processor are configured to transmit data via the integrated
memory; the central processor is configured for performing primary
noise reduction processing on a received electrocardiogram original
signal to obtain a primary electrocardiogram signal, and configured
for performing abnormal heartbeat classification preliminary
screening on characteristic data extracted by the graphic processor
to obtain suspected abnormal heartbeat data: and the graphic
processor is configured for performing characteristic extraction on
the primary electrocardiogram signal to obtain the characteristic
data, and configured for performing secondary noise reduction
processing on the primary electrocardiogram signal to obtain a
secondary electrocardiogram signal, and processing the suspected
abnormal heartbeat data and the secondary electrocardiogram signal
by using a template matching classification mode, to obtain final
abnormal heartbeat data.
2. The apparatus according to claim 1, wherein the central
processor is configured to perform the following operations:
receiving the electrocardiogram original signal and storing the
electrocardiogram original signal in the first memory; performing
primary noise reduction processing on the electrocardiogram
original signal loaded from the first memory, to obtain the primary
electrocardiogram signal, and storing the primary signal in the
second memory; and acquiring the characteristic data according to
storage location information, performing, in accordance with a set
rule decision mode, abnormal heartbeat classification on the
characteristic data to obtain suspected abnormal heartbeat data,
and storing the suspected abnormal heartbeat data in the second
memory.
3. The apparatus according to claim 2, wherein the graphic
processor is configured to perform the following operations:
performing characteristic extraction on the primary
electrocardiogram signal loaded from the second memory to obtain
characteristic data, and storing the characteristic data in the
second memory; performing secondary noise reduction processing on
the loaded primary electrocardiogram signal to obtain the secondary
electrocardiogram signal; and. acquiring the secondary
electrocardiogram signal from the second pre-process module to
obtain the suspected abnormal heartbeat data from the second
memory, reconfirming, in accordance with a set .sup.-template
matching mode and based on the secondary electrocardiogram signal,
the suspected abnormal heartbeat data to obtain the final abnormal
heartbeat data, and storing the final abnormal heartbeat data in
the second memory.
4. The apparatus according to claim 3, wherein the integrated
memory is further configured for mapping storage location
information of the characteristic data and the final abnormal
heartbeat data to the first memory, so as to enable the central
processor to acquire corresponding data according to the storage
location information.
5. The apparatus according to claim 3, wherein the operation of
performing primary noise reduction processing on the
electrocardiogram original signal loaded from the first memory to
obtain the primary electrocardiogram original signal comprises
performing filter processing on the electrocardiogram original
signal to obtain the primary electrocardiogram signal; and the
operation of performing secondary noise reduction processing on the
loaded primary electrocardiogram signal to obtain the secondary
electrocardiogram signal comprises performing artifact removal
processing on the primary electrocardiogram signal to obtain the
secondary electrocardiogram signal.
6. The apparatus according to claim 3, wherein the operation of
performing characteristic extraction on the primary
electrocardiogram signal loaded from the second memory to obtain
the characteristic data comprises: performing transformation on the
primary electrocardiogram signal, and outputting an
electrocardiogram signal in a morphological form; performing R wave
detection on the electrocardiogram signal in a morphological form,
and outputting a detection result; and performing QRS complex
detection on the detection result, and outputting characteristic
data containing QRS complex.
7. An electrocardiogram signal parallel analysis method, applied to
a mobile terminal, the mobile terminal comprising: an integrated
memory, a central processor and a graphic processor, wherein the
integrated memory is coupled to the central processor and to the
graphic processor and comprises a first memory configured for being
used by the central processor and a second memory configured for
being used by the graphic processor, the central processor is
configured to be able to access the second memory, and the central
processor and the graphic processor are configured to transmit data
via the integrated memory; and the method comprises: the central
processor performing primary noise reduction processing on a
received electrocardiogram original signal to obtain a primary
electrocardiogram signal; the graphic processor performing
characteristic extraction on the primary electrocardiogram signal
to obtain characteristic data; the central processor performing
abnormal heartbeat classification preliminary screening on the
chara.cteristic data to obtain suspected abnormal heartbeat data;
and the graphic processor performing secondary noise reduction
processing on the primary electrocardiogram signal to obtain a
secondary electrocardiogram signal, and processing the suspected
abnormal heartbeat data and the secondary electrocardiogram signal
by using a template matching classification mode .sup.-to obtain
final abnormal heartbeat data.
8. The method according to claim 7, wherein the method further
comprises: the central processor acquiring the final abnormal
heartbeat data, and uploading the final abnormal heartbeat data to
a remote medical platform; and the central processor receiving a
report fed back by the medical platform based on the final abnormal
heartbeat data.
9. The method according to claim 8, wherein a process of the
central processor and the graphic processor transmitting data via
the integrated memory comprises: the integrated memory copying, to
the second memory, data stored by the central processor in first
memory, and mapping, to the first memory, storage location
information stored by the graphic processor in the second
memory.
10. The method according to claim 7, wherein a process of the
central processor and the graphic processor transmitting data via
the integrated memory comprises: the integrated memory copying, to
the second memory, data stored by the central processor in first
memory, and mapping, to the first memory, storage location
information stored by the graphic processor in the second
memory.
11. A mobile terminal, comprising an electrocardiogram signal
parallel analysis apparatus, wherein the apparatus comprises an
integrated memory, a central processor and a graphic processor,
wherein the integrated memory is coupled to the central processor
and to the graphic processor and comprises a first memory
configured for being used by the central processor and a second
memory configured for being used by the graphic processor, the
central processor is configured to be able to access the second
memory, and the central processor and the graphic processor are
configured to transmit data via the integrated memory; the central
processor is configured for performing primary noise reduction
processing on a received electrocardiogram original signal to
obtain a primary electrocardiogram signal, and configured for
performing abnormal heartbeat classification preliminary screening
on characteristic data extracted by the graphic processor to obtain
suspected abnormal heartbeat data: and the graphic processor is
configured for performing characteristic extraction on the primary
electrocardiogram signal to obtain the characteristic data, and
configured for performing secondary noise reduction processing on
the primary electrocardiogram signal to obtain a secondary
electrocardiogram signal, and processing the suspected abnormal
heartbeat data and the secondary electrocardiogram signal by using
a template matching classification mode, to obtain final abnormal
heartbeat data.
12. The mobile terminal according to claim 11, wherein the central
processor is configured to perform the following operations:
receiving the electrocardiogram original signal and storing the
electrocardiogram original signal in the first memory; performing
primary noise reduction processing on the electrocardiogram
original signal loaded from the first memory, to obtain the primary
electrocardiogram signal, and storing the primary signal in the
second memory; and acquiring the characteristic data according to
storage location information, performing, in accordance with a set
rule decision mode, abnormal heartbeat classification on the
characteristic data to obtain suspected abnormal heartbeat data,
and storing the suspected abnormal heartbeat data in the second
memory.
13. The mobile terminal according to claim 12, wherein the graphic
processor is configured to perform the following operations:
performing characteristic extraction on the primary
electrocardiogram signal loaded from the second memory obtain
characteristic data, and storing the characteristic data in the
second memory; performing secondary noise reduction processing on
the loaded primary electrocardiogram signal to obtain the secondary
electrocardiogram signal; and acquiring the secondary
electrocardiogram signal from the second pre-process module to
obtain the suspected abnormal heartbeat data from the second
memory, reconfirming, in accordance with a set template matching
mode and based on the secondary electrocardiogram signal, the
suspected abnormal heartbeat data to obtain the final abnormal
heartbeat data, and storing the final abnormal heartbeat data in
the second memory.
14. The mobile terminal according to claim 13, wherein the
integrated memory is further configured for mapping storage
location information of the characteristic data and the final
abnormal heartbeat data to the first memory, so as to enable the
central processor to acquire corresponding data according to the
storage location information.
15. The mobile terminal according to claim 13, wherein the
operation of performing primary noise reduction processing on the
electrocardiogram original signal loaded from the first memory to
obtain the primary electrocardiogram original signal comprises
performing filter processing on the electrocardiogram original
signal to obtain the primary electrocardiogram signal; and the
operation of performing secondary noise reduction processing on the
loaded primary electrocardiogram signal to obtain the secondary
electrocardiogram signal comprises performing artifact removal
processing on the primary electrocardiogram signal to obtain the
secondary electrocardiogram signal.
16. The mobile terminal according to claim 13, wherein the
operation of performing characteristic extraction on the primary
electrocardiogram signal loaded from the secondary memory to obtain
the characteristic data comprises: performing transformation on the
primary electrocardiogram signal, and outputting an
electrocardiogram signal in a morphological form; performing R wave
detection on the electrocardiogram signal in a morphological form,
and outputting a detection result; and performing QRS complex
detection on the detection result, and outputting characteristic
data containing QRS complex.
Description
CROSS-REFERENCE TO RELATED APPLICATION
This application is a .sctn. 371 National Stage Application of
PCT/CN2017/113408, which was filed on Nov. 28, 2017, and claims
priority to the Chinese patent application with the filing No.
2017111696640, filed with the State Intellectual Property Office on
Nov. 21, 2017, and entitled "ECG Signal Parallel Analysis
Apparatus, Method and Mobile Terminal", content of which is
incorporated herein by reference in its entirety.
TECHNICAL FIELD
Present disclosure concerns to the technical field of
electrocardiography signal processing, particularly concerns to an
ECG signal parallel analysis apparatus, method and a mobile
terminal.
BACKGROUND ART
ECG (Electrocardiogram), which may display the evolution of cardiac
electrical activity over time, is one of the important
physiological data. Heart rate, rhythm disorders, or morphological
changes of electrocardiosignals may be pathological indicators. By
analyzing the recorded ECG waveform, myocardial infarction,
cardiomyopathy, myocarditis and various other heart diseases may be
detected.
In order to monitor long-term ECG signals, a high-performance
server is required to provide computing services. When a user
submits an enormous amount of electrocardiogram analysis requests
simultaneously in an unstable network environment, real-time
response is difficult for the traditional cloud platform-based ECG
signal analysis. If the analysis task of ECG signals is transferred
to the mobile terminal, due to the limited CPU (Central Processing
Unit) performance of the mobile terminal, it is still difficult to
handle long-term ECG signal processing and make timely feedback.
Meanwhile, since the processing needs to consume a large amount of
power of the apparatus, for a mobile terminal with limited battery
capacity, the battery losses are larger.
SUMMARY
In view of this, it is an object of the present disclosure to
provide an ECG signal parallel analysis apparatus, method and a
mobile terminal, to improve the analysis efficiency for ECG signals
so as to improve the timeliness of the analysis feedback of ECG
signals.
In order to achieve the above object, the technical solutions
adopted in the present disclosure are as follows.
In a first aspect, the present disclosure provides a parallel
analysis apparatus of an ECG signal, including: an integrated
memory, a CPU and a GPU, wherein the integrated memory includes a
first memory for being used by the CPU and a second memory for
being used by the GPU, and the CPU may access the second memory;
the CPU and the GPU are configured to transmit data via the
integrated memory; the CPU is used for performing primary noise
reduction processing on a received ECG original signal to obtain a
primary ECG signal, and used for performing abnormal heartbeat
classification preliminary screening process on characteristic data
extracted from the GPU to obtain suspected abnormal heartbeat data;
and the GPU is used for performing characteristic extraction on the
primary ECG signal to obtain the characteristic data, and used for
performing secondary noise reduction processing on the primary ECG
signal to obtain a secondary ECG signal, and processing the
suspected abnormal heartbeat data and the secondary ECG signal by
using a template matching classification mode, to obtain final
abnormal heartbeat data.
With reference to the first aspect, an embodiment of the present
disclosure provides a first possible example of the first aspect,
wherein the CPU includes an original signal reception module
configured for receiving an ECG original signal and storing the ECG
original signal in the first memory; a first pre-process module
configured for performing primary noise reduction processing on the
ECG original signal loaded from the first memory for primary noise
reduction processing to obtain a primary ECG signal, and storing
the primary ECG signal in the second memory; and a first abnormal
heartbeat classification module configured for acquiring
characteristic data according to storage location information,
performing, in accordance with a set rule decision mode, an
abnormal heartbeat classification on the characteristic data to
obtain suspected abnormal heartbeat data, and storing the suspected
abnormal heartbeat data in the second memory.
With reference to the first possible example of the first aspect,
an embodiment of the present disclosure provides a second possible
example of the first aspect, wherein the GPU includes: a
characteristic detection module configured for performing
characteristic extraction on the primary ECG signal loaded from the
second memory to obtain characteristic data, and storing the
characteristic data in the second memory; a second pre-process
module configured for performing secondary noise reduction
processing on the loaded primary ECG signal to obtain a secondary
ECG signal; and a second abnormal heartbeat classification module
configured for acquiring the secondary ECG signal from the second
pre-process module, acquiring the suspected abnormal heartbeat data
from the second memory, reconfirming, in accordance with a set
template matching mode and based on the secondary ECG signal, the
suspected abnormal heartbeat data to obtain final abnormal
heartbeat data, and storing the final abnormal heartbeat data in
the second memory.
With reference to the second possible example of the first aspect,
an embodiment of the present disclosure provides a third possible
example of the first aspect, wherein the integrated memory
includes: a mapping module, configured for mapping storage location
information of the characteristic data and the final abnormal
heartbeat data to the first memory, so as to enable the CPU to
acquire corresponding data according to the storage location
information.
With reference to the second possible example of the first aspect,
an embodiment of the present disclosure provides a fourth possible
example of the first aspect, wherein the first pre-process module
includes an IIR filter configured for performing filter processing
on the ECG original signal to obtain a primary ECG signal; and the
second pre-process module includes an artifact removal unit,
configured for performing an artifact removal process on the
primary ECG signal, to obtain a secondary ECG signal.
With reference to the second possible example of the first aspect,
an embodiment of the present disclosure provides a fifth possible
example of the first aspect, wherein the characteristic detection
module includes: a morphology transformation unit configured for
performing transformation on the primary ECG signal and outputting
an ECG signal in a morphological form; an R wave detection unit
configured for performing R wave detection on the ECG signal in a
morphological form and outputting a detection result; and a QRS
complex (QRS wave group) detection unit configured for performing
QRS complex detection on the detection result and outputting
characteristic data containing QRS complex.
In a second aspect, the present disclosure provides a method of
parallel analysis of an ECG signal, which is applied to a mobile
terminal, wherein the mobile terminal includes: an integrated
memory, a CPU and a GPU, the integrated memory includes a first
memory for being used by the CPU and a second memory for being used
by the GPU, and the CPU may access the second memory; the CPU and
the GPU are configured to transmit data via the integrated memory,
wherein the method includes: the CPU performing primary noise
reduction processing on a received ECG original signal to obtain a
primary ECG signal, the GPU performing characteristic extraction on
the primary ECG signal to obtain characteristic data, the CPU
performing abnormal heartbeat classification preliminary screening
process on the characteristic data to obtain suspected abnormal
heartbeat data, the GPU performing secondary noise reduction
processing on the primary ECG signal to obtain a secondary ECG
signal, and processing the suspected abnormal heartbeat data and
the secondary ECG signal by using a template matching
classification mode to obtain final abnormal heartbeat data.
With reference to the second aspect, an embodiment of the present
disclosure provides a first possible example of the second aspect,
wherein the method also includes: the CPU acquiring the final
abnormal heartbeat data and uploading the final abnormal heartbeat
data to a remote medical platform; and the CPU receiving a report
fed back by the medical platform based on the final abnormal
heartbeat data.
With reference to the second aspect or the first possible example
of the second aspect, an embodiment of the present disclosure
provides a second possible example of the second aspect, wherein a
process of the CPU and GPU transmitting data via the integrated
memory includes: the integrated memory copying the data stored by
the CPU in the first memory to the second memory, and mapping
storage location information of the data stored by the GPU in the
second memory to the first memory.
In a third aspect, the present disclosure provides a mobile
terminal that includes the above-described ECG signal parallel
analysis apparatus.
According to embodiments of the present disclosure, there are
provided an ECG signal parallel analysis apparatus, method, and a
mobile terminal, wherein the integrated memory includes a first
memory for being used by the CPU and a second memory for being used
by the GPU; the final abnormal heartbeat data is obtained by
performing, by the CPU, primary noise reduction processing on an
ECG original signal and performing an abnormal heartbeat
classification preliminary screening process on characteristic data
extracted from the GPU, performing, by the GPU, characteristic
extraction on the primary ECG signal obtained by the primary noise
reduction process, performing secondary noise reduction processing
on the primary ECG signal, and then processing suspected abnormal
heartbeat data obtained by the preliminary screening processing and
the secondary ECG signal obtained by the secondary noise reduction
process by applying a template matching classification mode. In
this method, the CPU and GPU co-process various tasks in the ECG
signal analysis process, the GPU completes complicated calculation
tasks in parallel processing mode, improving the analysis
efficiency for the ECG signal and thereby improving the timeliness
of the analysis and feedback of the ECG signal. Meanwhile,
equipment power consumption is reduced and user experience is
improved.
Further, the above way of data replication and mapping between
memories can avoid a large amount of data transmission and reduce
the data transmission time from the CPU side to the GPU side,
compared with the way of data transmission through buses and the
communication lines, further improving the efficiency of ECG signal
analysis.
Other features and advantages of the present disclosure will be set
forth in the following description, or, part of the features and
advantages may be inferred or undoubtedly determined from the
specification, or may be learned by implementing of above
techniques of the present disclosure.
The above objects, features and advantages of the present
disclosure will become more apparent from the following detailed
description of preferred embodiments thereof, taken in conjunction
with the accompanying figures.
BRIEF DESCRIPTION OF DRAWINGS
To describe the technical solutions in the embodiments of the
present disclosure or in the prior art more clearly, accompanying
figures required for describing the embodiments or the prior art
are introduced below briefly. Obviously, the accompanying figures
in the following description show some embodiments of the present
disclosure, and persons of ordinary skill in the art may still
derive other drawings from these accompanying drawings without
creative efforts.
FIG. 1 is a schematic structural diagram of an ECG signal parallel
analysis apparatus according to an embodiment of the present
disclosure;
FIG. 2 is a schematic structural diagram of another ECG signal
parallel analysis apparatus according to an embodiment of the
present disclosure;
FIG. 3 is a flow chart of an ECG signal parallel analysis method
according to an embodiment of the present disclosure; and
FIG. 4 is a flow chart of another ECG signal parallel analysis
method according to an embodiment of the present disclosure.
DETAILED DESCRIPTION OF EMBODIMENTS
To make the objectives, technical solutions, and advantages of the
disclosed embodiments more comprehensible, the technical solutions
of the present disclosure will be described clearly and completely
with reference to the accompanying figures. Apparently, the
described embodiments are only some of embodiments of the present
disclosure rather than all embodiments. All other embodiments
obtained by a person skilled in the art based on the embodiments of
the present disclosure without creative efforts shall fall within
the protection scope of the present disclosure.
Long-term ECG may be used to help diagnosis of heart diseases such
as intermittent cardiac arrhythmia. A user may acquire the ECG
signal through a wearable heart monitoring device and send the ECG
signal to a cloud platform connected with the device, with the
cloud platform analyzing and diagnosing the ECG signal and then
feeding back the diagnosis result to the monitoring device or the
user's mobile terminal. Due to volume of ECG signal data is large
and the cloud platform may continuously receive ECG signals sent
from a large number of users, the way of the cloud platform
processing the ECG signal poses a large computational burden to the
cloud processor, resulting in that the feedback timeliness and
reliability of ECG signal processing cannot be guaranteed.
In order to alleviate the above problems, the tasks of ECG signal
analysis and diagnosis can be completed by a mobile terminal such
as a wearable heart monitoring device, a mobile phone and a tablet
computer, etc.; however, due to the limited CPU performance of the
mobile terminal, it is still difficult to handle long-term ECG
signal processing and timely make feedback; at the same time, the
processing needs to consume a larger amount of power of the device,
and the battery loss is larger for a mobile terminal with limited
battery capacity.
In view of the problem of slow feedback of the ECG signal analysis
method described above, the embodiments of the present disclosure
provide an ECG signal parallel analysis apparatus, method and a
mobile terminal. The technology can be applied to wearable heart
monitoring devices, cell phones, tablet computers and the like
mobile terminals, and used in the scene of assisting the diagnosis
of intermittent cardiac arrhythmias and other heart diseases. The
technology can be implemented by related software or hardware, and
is described below by ways of embodiment.
Referring to a schematic structural diagram of an ECG signal
parallel analysis apparatus shown in FIG. 1, the apparatus
includes: an integrated memory 10, a CPU 11 and a Graphic
Processing Unit (GPU) 12. The integrated memory 10 includes a first
memory 101 for being used by the CPU 11 and a second memory 102 for
being used by the GPU 12, and the CPU 11 can access the second
memory 102. The CPU 11 and the GPU 12 transmit data via the
integrated memory 10.
The CPU 11 is used for performing primary noise reduction
processing on a received ECG original signal to obtain a primary
ECG signal; and used for performing abnormal heartbeat
classification preliminary screening on characteristic data
extracted from the GPU 12 to obtain suspected abnormal heartbeat
data.
The GPU 12 is used for performing characteristic extraction on the
primary ECG signal to obtain the characteristic data; and used for
performing secondary noise reduction processing on the primary ECG
signal to obtain a secondary ECG signal, and processing the
suspected abnormal heartbeat data and the secondary ECG signal by
using a template matching classification mode to obtain final
abnormal heartbeat data.
The above CPU and GPU are respectively used to perform different
tasks in ECG signal analysis, wherein some tasks may be performed
in parallel. For example, when the CPU performs an abnormal
heartbeat classification preliminary screening process, the GPU may
perform a secondary noise reduction process on the primary ECG
signal. Besides, due to the large amount of tasks undertaken by the
CPU, the execution of tasks with complicated computations of the
CPU is weak, and the GPU, which adopts a multi-core processing
manner, can handle in parallel the computationally complicated
tasks such as image computation, etc. or algorithms with inherent
parallel features. According to the attributes of the task, it can
improve the efficiency of ECG signal analysis by reasonably
distributing the processors for performing the task. For example,
the above characteristic extraction step, which usually requires
image recognition, calculation, etc. and requires a large amount of
computation, is performed by the GPU, which can greatly improve the
efficiency of ECG signal analysis.
Above integrated memory may be implemented by using an memory chip,
wherein the first memory and the second memory may be divided in a
form of software; the first memory can be used to store an ECG
original signal, a primary ECG signal, and suspected abnormal
heartbeat data and other data; and the second memory can be used to
store a primary ECG signal, characteristic data, suspected abnormal
heartbeat data, and final abnormal heartbeat data and other
data.
Data transmission between the first memory and the second memory
may be performed by copying and mapping. For example, the
integrated memory may copy, to the second memory, the data stored
by the CPU in the first memory and map, to the first memory, the
memory location information of the data stored by the GPU in the
second memory. Specifically, because the CPU can access both the
first memory and the second memory, when the CPU needs to acquire
the data in the second memory, only the storage address of the data
in the second memory needs to be mapped to the first memory, the
CPU accesses the second memory according to the mapped storage
address, and obtains corresponding data. Since the GPU can only
access the second memory, when the GPU needs to acquire the data in
the first memory, the integrated chip needs to copy the data from
the first memory into the second memory and then the data is read
by the GPU.
According to an embodiment of the present disclosure, there is
provided an ECG signal parallel analysis apparatus, wherein the
integrated memory includes a first memory for being used by the CPU
and a second memory for being used by the GPU. The final abnormal
heartbeat data is obtained by performing, by the CPU, primary noise
reduction processing on an ECG original signal and performing
abnormal heartbeat classification preliminary screening on
characteristic data extracted from the GPU, performing, by the GPU,
characteristic extraction on the primary ECG signal obtained by the
primary noise reduction process, performing secondary noise
reduction on the primary ECG signal, and processing suspected
abnormal heartbeat data obtained by the preliminary screening
process and the secondary ECG signal obtained by the secondary
noise reduction process by using a template matching classification
mode. In this manner, the CPU and GPU co-process various tasks in
the ECG signal analysis process, and the GPU completes complicated
calculation tasks in parallel processing mode, improving the
analysis efficiency for the ECG signal and thereby improving the
timeliness of the analysis and feedback of the ECG signal.
Meanwhile, equipment power consumption is reduced and user
experience is improved.
Further, the above way of data replication and mapping between
memories can avoid a large amount of data transmission and reduce
the data transmission time from the CPU side to the GPU side,
compared with the way of data transmission through the buses and
the communication lines, further improving the efficiency of ECG
signal analysis.
Referring to the schematic structural diagram of another ECG signal
parallel analysis apparatus shown in FIG. 2, the apparatus is
implemented on the basis of the apparatus shown in FIG. 1. The
apparatus includes an integrated memory 10, a CPU 11 and a GPU 12.
The integrated memory 10 includes the first memory 101 for being
used by the CPU 11, and the second memory 102 for being used by the
GPU 12, and the CPU 11 can access the second memory 102. The CPU 11
and the GPU 12 transmit data via the integrated memory 10.
Mobile terminals such as smartphones have highly integrated
circuits that combine major components (such as CPU, GPU, memory,
etc.) into a single chip. This approach enables high-bandwidth data
transmission; and at the same time, the ultra-bandwidth memory
indicator accelerates the speed of data transmission between the
memory and the CPU/GPU. In addition, the CPU and GPU memories are
integrated on the same chip, and separated by embedded software.
Tasks are transferred during task execution, so memory-mapping
techniques can be introduced to map the same piece of physical
memory into the memory spaces of the CPU and GPU to reduce or even
avoid data transmission.
The CPU specifically includes: an original signal reception module
111 configured for receiving an ECG original signal and storing the
ECG original signal in the first memory; a first pre-process module
112 configured for performing primary noise reduction processing on
the ECG original signal loaded from the first memory for to obtain
a primary ECG signal, and storing the primary ECG signal in the
second memory; a first abnormal heartbeat classification module 113
configured for acquiring characteristic data according to storage
location information, performing, in accordance with a set rule
decision mode, abnormal heartbeat classification on characteristic
data to obtain suspected abnormal heartbeat data, and storing
suspected abnormal heartbeat data in the second memory.
The original signal reception module can be connected with an
electrocardiogram sensor; and the electrocardiogram sensor can
sense the action potential waveform of cells in different regions
of the heart and convert it into a signal that can be output,
wherein the signal is the ECG original signal.
The first pre-process module may include an Infinite Impulse
Response (IIR) filter configured for filter processing the ECG
original signal to obtain a primary ECG signal. Of course, the
filter processing may also be implemented by other filters, such as
a Finite Impulse Response (FIR) filter. Due to the tightly coupled
mode of the IIR filter, parallelization is difficult to achieve, so
the IIR filter is implemented in the CPU. After the first
pre-process module processes and obtains the primary ECG signal,
the primary ECG signal is usually first stored in the first memory;
and because the subsequent characteristic extraction process is
performed by the GPU, the integrated memory copies the primary ECG
signal to the second memory, for acquisition by the GPU.
The first abnormal heartbeat classification module may acquire a
predefined rule determination mode from the first memory. The rule
determination mode may be implemented as a parameter threshold. For
example, if one or more parameters in the characteristic data are
greater than corresponding threshold values, it can be initially
determined that there is abnormality of the ECG signal; the type of
the abnormality may also be preliminarily classified according to
the threshold values to obtain the suspected abnormal heartbeat
data, and then the suspected abnormal heartbeat data may be saved
again. The suspected abnormal heartbeat data processed and obtained
by the first abnormal heartbeat classification module is usually
firstly stored into the first memory; and since the subsequent
re-confirmation processing of the suspected abnormal heartbeat data
is performed by the GPU, the integrated memory copies the
classification results to the second memory for acquisition by the
GPU.
The GPU specifically includes: a characteristic detection module
121 configured for performing characteristic extraction on the
primary ECG signal loaded from the second memory to obtain
characteristic data, and storing the characteristic data in the
second memory; a second pre-process module 122 configured for
performing secondary noise reduction processing on the loaded
primary ECG signal to obtain a secondary ECG signal; a second
abnormal heartbeat classification module 123 configured for
acquiring the secondary ECG signal from the second pre-process
module, acquiring suspected abnormal heartbeat data from the second
memory, reconfirming, in accordance with a set template matching
mode and based on the secondary ECG signal, the suspected abnormal
heartbeat data to obtain final abnormal heartbeat data, and storing
the final abnormal heartbeat data in the second memory.
The above-mentioned characteristic detection module can be
implemented by various characteristic extraction algorithms, such
as machine learning, wavelet transformation, morphological
transformation, etc. In view of the particularity of the ECG
signal, in order to balance the accuracy and high efficiency of ECG
signal characteristic recognition, the present embodiment is
specifically implemented in the following manner: specifically, the
characteristic detection module includes: (1) a morphology
transformation unit, configured for performing transformation on
the primary ECG signal, and outputting an ECG signal in
morphological form; (2) an R wave detection unit, configured for
performing R wave detection on the ECG signal in a morphological
form, and outputting a detection result; and (3) a QRS complex
detection unit, configured for performing QRS complex detection on
the detection result, and outputting characteristic data containing
QRS complex.
In the ECG signal, an R wave is a positive wave firstly appears in
a signal period and located above a reference horizontal line. The
QRS complex includes an R wave, a Q wave, an S wave, an R' wave, an
S' wave and a QS wave. By detecting parameters of width, time
internal, amplitude, shape and so on of these waveforms, a variety
of characteristic data may be obtained.
The characteristic data extracted by the characteristic detection
module is usually stored in the second memory. Although the CPU can
access the second memory, position of the data stored in the second
memory needs to be mapped to the first memory. Based on this, the
integrated memory includes a mapping module 103 configured for
mapping storage location information of the characteristic data and
the final abnormal heartbeat data to the first memory, so as to
enable the CPU to acquire corresponding data according to a storage
location information. This manner makes the CPU be able to more
quickly obtain the data processed and obtained by the GPU, avoiding
the time-consuming data transmission, thereby improving the
analysis efficiency for the ECG signal.
The second pre-process module may include an artifact removal unit
configured for performing artifact removal on a primary ECG signal
to obtain a secondary ECG signal. In general, the signal detected
by a sensor from a body surface electrode contains different types
of interference, such as power frequency interference, baseline
drift, electrode contact noise, electromyography interference, and
movement interference and so on, these interference forming
artifacts in the ECG signal. In order to obtain relatively pure ECG
signals so as to improve the accuracy of subsequent characteristic
detection and heartbeat abnormity recognition, the present
embodiment adopts above artifact removal unit to perform artifact
removal processing on the primary ECG signal.
The second abnormal heartbeat classification module acquires the
secondary ECG signal from the second pre-process module on one hand
and acquires the suspected abnormal heartbeat data from the second
memory on the other hand, and the suspected abnormal heartbeat data
is in advance copied from the first memory to the second memory;
the second abnormal heartbeat classification module generates a QRS
standard template from a secondary ECG signal of a noise-free
signal, and then corrects the data of erroneous determination of
the suspected abnormal heartbeat data according to the standard
template to generate final abnormal heartbeat data. The final
abnormal heartbeat data is saved to the second memory, and the
address of the final abnormal heartbeat data in the second memory
is mapped to the first memory for acquisition by the CPU. After the
CPU acquires the final abnormal heartbeat data, the data may be
pushed to a user terminal, uploaded to a cloud platform, or
subjected to other processing.
Besides, the ECG signal parallel analysis apparatus may also be
further optimized through workgroup size, data vectorization
operation and zero memory copy technology, improving the efficiency
of the analysis.
In the above manners, the CPU and the GPU co-process various tasks
in ECG signal analysis process, and the GPU complete complex
computing tasks in a parallel processing manner, improving the
efficiency of ECG signal analysis, thereby improving the timeliness
of ECG signal analysis and feedback. Meanwhile, the device power
consumption is reduced, and user experience is improved.
Corresponding to the embodiments of the apparatuses, referring to a
flow chart of an ECG signal parallel analysis method shown in FIG.
3, the method is applied to a mobile terminal. The mobile terminal
includes: an integrated memory, a CPU and a GPU, wherein the
integrated memory includes a first memory for being used by the CPU
and a second memory for being used by the GPU, and the CPU may
access the second memory; and the CPU and the GPU transmit data via
the integrated memory. The method includes the following steps:
Step S302, the CPU performing primary noise reduction processing on
a received ECG original signal to obtain a primary ECG signal;
Step S304, the GPU performing characteristic extraction on the
primary ECG signal to obtain characteristic data;
Step S306, the CPU performing abnormal heartbeat classification
preliminary screening on the characteristic data to obtain
suspected abnormal heartbeat data; and
Step S308, the GPU performing secondary noise reduction processing
on the primary ECG signal to obtain a secondary ECG signal, and
processing the suspected abnormal heartbeat data and the secondary
ECG signal by using a template matching classification mode, to
obtain final abnormal heartbeat data.
In the ECG signal parallel analysis method according to the
embodiment of the present disclosure, the CPU performs primary
noise reduction processing on the ECG original signal, and the GPU
performs characteristic extraction on the primary ECG signal
obtained by the primary noise reduction processing; the CPU
performs abnormal heartbeat classification preliminary screening on
extracted characteristic data, and the GPU performs secondary noise
reduction processing on the primary ECG signal, and then processes,
by applying a template matching classification mode, suspected
abnormal heartbeat data obtained by the preliminary screening and a
secondary ECG signal obtained by the secondary noise reduction, to
obtain final abnormal heartbeat data. In this manner, the CPU and
the GPU co-process various tasks in the ECG signal analysis
process, and the GPU completes complicated calculation tasks in
parallel processing mode, improving the analysis efficiency of the
ECG signal and thereby improving the timeliness of the analysis and
feedback of the ECG signal, and meanwhile reducing equipment power
consumption and improving user experience.
Referring to a flow chart of another ECG signal parallel analysis
method shown in FIG. 4, this method is implemented based on the
method shown in FIG. 3. The method is implemented by multi-party
interaction between a CPU of a mobile terminal, a first memory and
a second memory in an integrated memory, and a GPU, wherein the
second memory may also be referred to as a video memory. The method
includes the following steps:
Step S402, the CPU receives an ECG original signal, and stores the
ECG original signal in the first memory;
Step S404, the CPU loads the ECG original signal in the first
memory for primary noise reduction processing to obtain a primary
ECG signal;
Step S406, the CPU stores the primary ECG signal in the first
memory;
Step S408, the integrated memory copies to a second memory the
primary ECG signal stored by the CPU in the first memory;
Step S410, the GPU loads the primary ECG signal from the second
memory for characteristic extraction to obtain characteristic
data;
Step S412, the GPU stores the characteristic data in the second
memory;
Step S414, the integrated memory maps to the first memory storage
location information of the characteristic data stored by the GPU
in a second memory;
Step S416, the CPU acquires the characteristic data according to
the storage location information, and performs, in accordance with
a set rule decision mode, abnormal heartbeat classification on the
characteristic data to obtain suspected abnormal heartbeat
data;
Step S418, the CPU stores the suspected abnormal heartbeat data in
the first memory;
Step S420, the integrated memory copies, to the second memory,
suspected abnormal heartbeat data stored by the CPU in the first
memory;
Step S422, the GPU loads the primary ECG signal for secondary noise
reduction processing, to obtain a secondary ECG signal; and in
order to make full use of the heterogeneous computing resources of
the CPU and the GPU, in this method, the secondary noise reduction
process for removing artifacts of the ECG signal is adjusted from
before the characteristic extraction to before the abnormal
heartbeat re-confirmation;
Step S424, the GPU acquires the suspected abnormal heartbeat data
from the second memory, and reconfirming, in accordance with a set
template matching mode and based on the secondary ECG signal, the
suspected abnormal heartbeat data, to obtain final abnormal
heartbeat data;
Step S426, the GPU stores the final abnormal heartbeat data in the
second memory;
Step S428, the integrated memory maps, to the first memory, storage
location information of the final abnormal heartbeat data stored by
the GPU in the second memory;
Step S430, the CPU acquires the final abnormal heartbeat data;
Step S432, the final abnormal heartbeat data is uploaded to a
remote medical platform, to enable the medical platform to generate
a feedback report according to the final abnormal heartbeat data;
and
Step S434, the CPU receives the feedback report based on the final
abnormal heartbeat data from the medical platform.
In the above manner, the CPU and the GPU co-process various tasks
in the ECG signal analysis process, and the GPU completes
complicated calculation tasks in parallel processing mode,
improving the analysis efficiency for the ECG signal and thus
improving the timeliness of the analysis and feedback of the ECG
signal, and meanwhile reducing equipment power consumption and
improving user experience.
Corresponding to the above apparatus and method embodiments, an
embodiment of the present disclosure further provides a mobile
terminal, which includes the above ECG signal parallel analysis
apparatus.
The ECG signal parallel analysis apparatus, method and mobile
terminal provided by the embodiments of the present disclosure
propose a new automatic ECG parallel analysis manner based on the
Mobile Graphics processing unit (GPU), wherein compared with the
sequential analysis manner of ECG signals, in the parallel manner,
the whole program flow is reorganize and the CPU/GPU heterogeneous
computing resources are fully utilized. This manner can
significantly shorten the ECG data execution time of 24 hours,
wherein through optimization in aspects of data vectorization, work
group resizing and zero memory copy and others, the above execution
time is further reduced, the feedback efficiency is improved, and
the user experience is improved. Besides, when a large amount of
computation is distributed to the GPU, the average power
consumption of the test mobile device is small, alleviating the
problem of limited battery working life of the mobile device.
In the several embodiments provided in the present disclosure, it
should be understood that the disclosed apparatuses and methods may
also be implemented in other manners. The apparatus embodiments
described above are merely illustrative, for example, the flowchart
and block diagrams in the figures illustrate the system structures,
functions, and operations of possible implementations of
apparatuses, methods and computer program products according to
various embodiments of the present disclosure. In this regard, each
block of the flowcharts or block diagrams may represent a module, a
section of a program, or a portion of a code, and the module,
section of a program, or portion of a code includes one or more
executable instructions for implementing the specified logic
functions. It should also be noted that in some alternative
implementations, the functions marked in the blocks may occur in an
order different from that marked in the figures. For example, two
consecutive blocks may in fact be executed substantially in
parallel, and sometimes they may be executed in the reverse order,
depending on the function involved. It is also to be noted that
each block of the block diagrams and/or flowcharts, and
combinations of blocks in the block diagrams and/or flowcharts can
be implemented by special hardware-based systems that perform the
specified functions or actions, or can be implemented by a
combination of dedicated hardware and computer instructions.
Besides, each function module or unit in the embodiments of the
present disclosure may be integrated together to form an
independent part, or the modules may exist separately, or two or
more modules may be integrated to form an independent part.
The function, if implemented in the form of a software functional
unit and sold or used as a separate product, may be stored in a
computer-readable storage medium. Based on this understanding, the
technical solutions of the present disclosure essentially, or the
contributing parts to the prior art, or part of the technical
solutions may be embodied as a software product, with the computer
software product stored in a storage medium and including several
instructions used to enable a computer device (may be a personal
computer, a server, or a network device, etc.) to execute all or
part of the steps of the method according to the embodiments of the
present disclosure. The foregoing storage medium includes various
media capable of storing program codes, such as a USB flash disk, a
removable hard disk, a read-only memory (ROM), a random access
memory (RAM), a magnetic disk, or an optical disk, etc.
Finally it should be noted that the above embodiments are merely
specific implementations of the present disclosure for illustrating
the technical solutions of the present disclosure rather than
limiting the present disclosure, the protection scope of the
present disclosure is not limited thereto, although the present
disclosure has been described in detail with reference to the
foregoing embodiments, those skilled in the art should understand:
anyone skilled in the art may still make modifications to the
technical solutions described in the foregoing embodiments or
easily conceivable variations within the technical scope disclosed
in the present disclosure, or replace some of the technical
features equivalently; yet these modifications, variations or
substitutions do not make the essence of the corresponding
technical solutions depart from the spirit and scope of the
technical solutions of the embodiments of the present disclosure,
and should all fall within the protection scope of the present
disclosure. Therefore, the protection scope of the present
disclosure should be subject to the protection scope of the
claims.
* * * * *